Complementarity divides different kinds of information

What it is

Complementarity means each channel carries a different part of the message, and together they form a complete expression. Speech states the action and pointing names the object; vision shows state and haptics confirm receipt. Complementarity does not duplicate content, so it carries no integration burden—but it requires users to combine the parts correctly.

Why it happens

Its efficiency comes from division of labor: each channel takes what it does best, with speech for meaning, pointing for reference, and vision for spatial relations. Each channel then stays within its capacity, and the total exceeds any single channel. The price is a combination requirement: users must recognize that the parts belong to one expression rather than two independent events. That combination depends on temporal proximity and semantic compatibility; without them, complementarity degrades into two unrelated events.

Studying it

Measure combination success: build tasks that require combining information across channels and record whether users combine correctly, how long it takes, and what errors occur. Variables include the interval between parts, semantic compatibility, and user experience. Outcomes include combination accuracy and delay, the latter reflecting combination cost. Include a pure redundancy condition as a control to show the different cost structures.

Where it stops holding

When the two parts cannot be clearly linked—too long an interval, mismatched semantics—complementarity fails and produces errors. If a user lacks sensory access to one channel, part of the complement is missing and the expression is incomplete, so each channel needs a substitute path. When the task needs only one piece of information, complementarity has no purpose.

Applying it

• Assign information types by channel strength so each carries what it does best.
• Keep the parts close in time and semantically combinable so users can tell they belong together.
• Keep a substitute path for each channel so no channel's absence makes the whole incomplete.
• Verification: measure accuracy and time for combining the parts and compare against a redundant scheme to confirm the combination cost is acceptable.

Related

• Within the group: D5.01.1 Redundancy improves reliability but adds load · D5.01.3 Choosing complementarity or redundancy depends on environmental uncertainty
• Adjacent: D5.02.1 Speech supplies the action, pointing supplies the object · D4.04.1 Multichannel information should point at one interpretation
• Search terms: complementarity · modal division of labor · multimodal input